Thanks to visit codestin.com
Credit goes to github.com

Skip to content
This repository was archived by the owner on Feb 25, 2025. It is now read-only.

Conversation

@Phil26AT
Copy link
Collaborator

@Phil26AT Phil26AT commented Apr 6, 2022

  • Adds CUDA support when COLMAP was built with CUDA_ENABLED
    • use pycolmap.has_cuda in python to check for GPU support
    • device argument in functions to specify whether to use CPU ("cpu"), GPU ("cuda"), or "auto" for GPU if available else CPU.
    • all functions except for patch_match_stereo can be used when built w/o CUDA_ENABLED.
  • Reorganized pipeline functions (see pipeline/)
  • SIFT feature extractor and matcher (exhaustive, sequential, spatial, vocabtree) + options
  • Dense MVS reconstruction pipeline (undistort_images,patch_match_stereo, stereo_fusion) + options
  • SIFT GPU on numpy image (see sift.cc), both byte and float images supported
  • Allow interrupting long c++ functions via KeyboardInterrupt

@sarlinpe sarlinpe merged commit b974ced into master Apr 20, 2022
@sarlinpe sarlinpe deleted the cuda-bindings branch April 20, 2022 21:33
@julien-blanchon
Copy link

Did you have a prebuild weel with cuda ?

@sarlinpe
Copy link
Member

sarlinpe commented Sep 27, 2023

We don't but any help is very much appreciated. Some pointers:

  • We'd need to install the Cuda toolkit in the CenOS manylinux image, either through yum (here is a good starting point) or via pip (described here)
  • It would be nice to not package the Cuda libs in the wheel and instead depend on Cuda pip packages - we'd need to add these requirements to the pycolmap wheels.
  • I don't know whether we need to build a separate wheel for each version of Cuda

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants